Increased conflicts between timber production and environmental protection led some analysts to advocate land-use segregation, often referred to as forest management zoning. The objective of zoning is to create ecologically desirable non-fragmented forest reserves and group timber production areas. We formulate an integer programming model of forest zoning that explicitly addresses clustering of spatial units allocated to timber production and reserve zones while also promoting separation of these zones. A tabu search algorithm is developed, implemented and tested using a case study. The case study results indicate that up to 5% of the net financial return is sacrificed with a 'satisfactory' grouping of units within each zone. A 'good' separation between the reserves and timber production zone is achieved at the cost of further decline of the net financial return up to 11% relative to the unconstrained case.